For BYU, 1960 was a watershed year: BYU Studies began its first full year of publication.1 KBYU–FM aired its first broadcast. The Ballroom Dance Company was formed. Three BYU students formed the Lettermen and recorded their first song.2 The honors program was established. The MBA program was approved. Rex Lee, future BYU president, was student body president. Enrollment topped ten thousand for the second year in a row (up from forty-five hundred in 1950). In the midst of construction in Provo that would more than double the value of its physical facilities, BYU also purchased 135 acres in Anaheim, California;3 313 acres in Portland, Oregon; and 249 acres in Phoenix, Arizona,4 as sites for satellite campuses in an ambitious expansion program.5 And a statistics department was organized at BYU.
Of course, the campuses in Anaheim, Portland, and Phoenix were never built, and the formation of a statistics department with just one professor and five students seems hardly worth mentioning, given concurrent major developments.6 But these events are connected in a fascinating way. The Statistics Department slipped into existence through the chance opening of a narrow window of opportunity directly tied to the confidence of the 1950s, a rapidly growing church, escalating enrollment at BYU, and BYU’s land purchases. The interconnections of these factors reveal much about a remarkable historical period for both the Church and BYU.
Demand for Wood and a Statistics Department
Keenly aware of the rapidly increasing enrollment at BYU, and with even greater increases looming as baby boomers reached university age in the 1960s and 1970s, BYU President Ernest L. Wilkinson formed the Bureau of Church Studies at BYU in the mid-1950s to predict overall Church membership and, more specifically, university-aged Church membership through the year 2000.7 To spearhead the study, he hired an academic from the Stanford Research Institute, Howard Nielson,8 whose work had caught his attention: Nielson had created for the Weyerhaeuser Timber Company a sophisticated projection entitled America’s Demand for Wood (see fig. 1).9 At the time he contacted Nielson, Wilkinson did not know Nielson was a member of the Church, born in Richfield, Utah, and educated at both BYU and the University of Utah.10
Nielson completed the Church growth study (discussed in the next section) in 1957, and the results provided Wilkinson with the forecasts he needed to facilitate long-term planning for the Church Educational System.11 When he discovered Nielson was LDS, Wilkinson urged him to stay at BYU teaching statistics as part of the Economics Department. Nielson decided to stay for a year, ultimately teaching statistics classes in four different departments: economics, agricultural economics, accounting, and mathematics.12 Among other difficulties associated with working for several departments, in the winter term of 1958 his name was spelled four different ways in the class schedule.
After Nielson’s first year at BYU, he was offered a job by IBM at three times his BYU salary, based on the continued esteem of the America’s Demand for Wood study.13 He talked to Dean Weldon Taylor about the decision he was facing.14 In the course of the conversation, Nielson mentioned that teaching statistics at BYU would be a lot easier if one department, a statistics department, did all the teaching. Taylor, apparently fearing that Nielson would take the IBM job, said he would form the department. Nielson agreed to stay; he wanted to raise his growing family in the friendly atmosphere of Provo and thought he could make up much of the difference in salary by consulting with Hill Air Force Base and Hercules Powder Company.15 Nielson later admitted that his suggestion for a statistics department was a long shot; he did not anticipate Taylor’s expeditious support and would likely have stayed even without a statistics department to sweeten the deal.16 Thus an unintended consequence of the 1957 membership projection for the rapidly growing Church was the formation of the Statistics Department at BYU (see fig. 2).
The almost nonchalant decision to form a statistics department in order to retain a promising faculty member says something about the organizational climate of BYU at the time. It also was a larger academic step than the administration may have realized. Statistics was a latecomer as a distinct academic discipline. The first statistics departments in the United States were formed as recently as 1934 at Iowa State University, 1935 at George Washington University, and 1941 at North Carolina State College.17 Later, in the 1940s and 1950s, amid a perfect storm of postwar science, computing, and population growth, departments sprang up at about a dozen research universities, including Stanford, Cornell, Berkeley, and Harvard. However, such eminent universities as Wisconsin, Texas A&M, and Yale added statistics departments in the 1960s, after the formation of BYU’s department.
Even today, a few major universities do not have distinct departments of statistics; neither the University of Utah nor Utah State University currently has a statistics department.18 So having a statistics department at BYU in 1960 was, if not quite avant-garde, decidedly progressive and unique for a school not in the same league as major research universities. The first PhD program at BYU had been started only three years earlier; it was not even clear at the time that there would be trained LDS statisticians available to staff the new department.
The 1957 Church Membership Projection
As indicated, establishing a statistics department at BYU was a direct, if perhaps fortuitous, result of the Church membership projection study that President Wilkinson hired Howard Nielson to spearhead. Nielson’s 1954 report America’s Demand for Wood was a careful projection of future demand for wood based on assumptions about increasing demand, competition, technological changes, and shifts in home design.19 The forecast received wide acclaim, and various sectors of the economy, including education, grasped the utility of such forecasts amid the rapid growth of the time. Wilkinson desired a similar projection for Church membership from 1957 to 2000.
The 1957 Church membership study identified six regions expected to have more than 5,000 LDS members of college age by 1975 and at least 10,000 by 2000. It identified twenty additional regions that were expected to have at least 2,000 members of college age by 1975. Based on these projections, Wilkinson proposed capping enrollment at BYU in Provo at 12,000–15,000 and building up to ten additional colleges in the Church system. The executive board supported Wilkinson’s vision of BYU campuses serving LDS youth in areas of membership concentration throughout the western United States.
In the first step to implement the concept, plans were announced to move Ricks College from Rexburg to Idaho Falls.20 By 1960, President David O. McKay had changed Wilkinson’s title in the Church Unified School System from Administrator21 to Chancellor—reflecting Wilkinson’s position as head of a university system, not just a single university—and eight million dollars had been spent to purchase 1,650 acres for Church colleges in Anaheim, San Fernando, Hayward (San Francisco area), Phoenix, Portland, Idaho Falls, and Salt Lake City. Almost as quickly as it began, however, the expansion concept stalled. A concerted campaign in Rexburg led to a reversal of the decision to move Ricks College,22 and financial hesitation on the part of the First Presidency at the expenses of constructing the other new campuses (fortunate, as it turned out) led to the abandonment of the whole expansion project.23
The membership projection, however, is interesting in its own right. The report was daring. Even now, few statisticians would feel comfortable predicting values forty-three years in the future. Those who would dare to make such a prediction might hope that they would not be around when someone inevitably compares their predictions to actual realizations as we do here.24 Some long-range predictions were amazingly accurate, while some were off by orders of magnitude. The accurate predictions attest to the validity and robustness of the modeling methods and assumptions implemented via hand computations. The wildly inaccurate predictions give a glimpse into unexpected changes in migration patterns within the United States and, much more importantly, unexpected Church expansion outside of the United States and Canada.25
The membership projection was not simply an extrapolation of the overall trend. It took into account the age structure of the Church in 1957, age-specific birth and death rates, conversion rates, geographical differences, and immigration patterns (see fig. 3). It made conservative assumptions about how these demographic characteristics would change in the future. The report divided the Church into forty-five regions, most in the western United States and Canada, producing a separate projection for each region.
Interesting facts brought out in the study include the following: all but two of the forty-five regions had contributed to the growth of the Church in the previous fifteen years; Church membership in twenty-eight of the regions had more than doubled; overall Church membership grew by a healthy 27.4 percent between 1950 and 1957; and, amazingly, the San Francisco, Phoenix, and Los Angeles regions each had a membership growth rate of around 80 percent in that seven-year period.
Not only was membership growing, but it was growing at a faster rate than the rapidly growing general population. Only southern Utah and Las Vegas had a smaller percentage of Latter-day Saints in 1957 than in 1910. Of interest to statisticians is a paradoxical finding that Arizona was the only western state in which LDS membership did not grow as a percentage of total population in the state—despite the fact that the LDS percentage increased in every area within the state.26
Overall, the membership prediction for 2000 was off by almost 80 percent; it predicted total Church membership in 2000 would be 6.7 million, while in actuality it reached 11.1 million (see fig. 4). At the time, even a projection of 6.7 million seemed unrealistically optimistic. Nielson compared this projection to an earlier forecast of 3.6 million27 for the year 2000 and felt compelled to comment that even though his projection seemed unrealistic, it was justified by assumptions that were, if anything, conservative.
While the accuracy of the overall projection is not impressive, many of the regional projections for the United States and Canada were remarkably accurate (see fig. 5). There were two large but offsetting exceptions to accuracy for the U.S. and Canadian projections: membership in California was overestimated by around one million members, and membership in the nonwestern states and provinces was underestimated by almost one million members. Thus, the Church grew in the United States as a whole as predicted, but migration (Church members and otherwise) to the southern states was much greater than expected, while migration to California was much lower than expected.
Census Bureau statistics indicate that between 1940 and 1950, population growth rates for California, Utah, and the southern states were 53 percent, 25 percent, and 13 percent, respectively.28 Between 1990 and 2000, the comparable figures were 15 percent, 30 percent, and 18 percent. Nielson’s model assumed that extensive migration to California would persist. Based on that assumption, the 1957 report projected that by 1985 Los Angeles would surpass Salt Lake City as the largest membership region in the Church; San Francisco would be third largest, followed by the southern states, Ogden, and Provo.
The most glaringly inaccurate projections in the 1957 report involved Mexico, Latin America, Asia, Africa, and eastern Europe. The model obviously could not have anticipated the 1978 declaration extending the priesthood to all worthy male members29 or the demise of the Soviet Empire. But even adjusting for those events, Church growth in most of these regions could not have been predicted from past trends in 1957. The projection is therefore a staggering commentary on the miraculous and statistically unexpected growth of the Church in these areas.
The 1957 model assumed the number of new converts would increase by 1,000 per year until 1975, with half that annual increase after 1975. In actuality, the missionary force grew at a much faster rate than predicted, as did the number of converts baptized (see fig. 6).30
The End of the Junior College Proposal
In March 1963, the executive committee of the Church Unified School System voted to abandon the expanded college proposal.31 Wilkinson later revised his proposal, reducing the number of junior colleges from ten to four: a school in Mexico City to train CES teachers for Mexico, Ricks College, a junior college in Anaheim, and a junior college in Phoenix with special emphasis on Native American students. He suggested that these schools would more than pay for themselves because of the increased tithing the Church would ultimately receive from students acquiring complete religious education at junior colleges. Even so, the proposal was shelved and many of the properties were sold at several times the acquisition price.32
Wilkinson said he spent more time on the junior college proposal than on anything else in his first twelve years at BYU. His eleven-month absence from BYU after resigning in 1964 to run for U.S. Senate proscribed any return to the junior college proposal. While his vision of a constellation of BYU campuses was never realized, Wilkinson’s tenure at BYU saw an increase in enrollment from 4,004 students in February 1951 to 25,116 students in September 1971, one month after his retirement; this 527 percent growth dwarfed the national average among universities of 75 percent in the same period.33
The hesitation of the First Presidency to build the satellite campuses indicates that they were not comfortable with Nielson’s careful statistical projection. Their lack of confidence in the projection was well justified. The Church did not grow as dramatically as expected in California, and it grew much more dramatically than expected outside of the United States and Canada. Had the campuses been built as suggested by the 1957 projection, they would not have been located where they would have best served the youth of the Church. Also, needed resources for churches and temples outside of the United States and Canada would have been reduced. In retrospect, a central large BYU fits the needs of the international Church better than satellite campuses based on a very careful 1957 projection. One unintended benefit of President Wilkinson’s attempt to project Church growth, however, is that BYU now has a thriving Department of Statistics, the only one in the state of Utah.
Fifty Years of the BYU Department of Statistics
In addition to its remarkable origin connected with Howard Nielson’s statistical projections, many things about the new Statistics Department at BYU in 1960 were unique. The department initially offered only a bachelor’s degree even though most existing departments exclusively offered graduate degrees. As late as 1986, the president of the American Statistical Association (ASA), possibly as a result of a recent visit to BYU, was urging statistics departments to develop bachelor’s programs.
The founder of the department, Howard Nielson, had a PhD in business with a minor in statistics. The next two faculty members hired into the department also had minors in statistics, both from North Carolina State College.34 Mel Carter, who had been working as a statistical consultant at Purdue and North Carolina State College, had a PhD in animal nutrition, and Gill Hilton had a PhD in soil science.
Carter brought with him rough-hewn qualities that defined the Statistics Department in its early years. Carter had intended to work at his family dairy farm for life, but after a disagreement with his older brother about who was doing more work, he left for graduate study. His laboratory experiences taught him that influences as subtle as alfalfa dust could completely disrupt a nutrition experiment, but statistical design and analysis tools were available to adjust for these subtle effects. So he learned as much about statistics as possible from pioneers in the field, including Gertrude Cox. Carter jumped at the invitation to join the BYU department even though he suspected that Nielson “probably didn’t know that he wasn’t a statistician.” Carter’s unique combination of humility, rough edges, and respect for statistical models drew students to him and the major.
Gill Hilton brought with him a similar respect for the power of statistics in applied science combined with a strong administrative ability. Hilton later served as department chairman for twelve years, which proved critical to the growth of the department. After his service, he continued his administrative career as a mission president in Virginia.
By 1963, Nielson had built the department up to five faculty members by hiring Earl Faulkner, a biostatistics PhD from the University of Minnesota, and Dale Richards, an operations research PhD from the Department of Industrial Engineering at Iowa State.35 A graduate program for the department was approved and the curriculum was stable when Carter became department chair. Under Carter’s leadership (1963–66), the new graduate program was developed and a tradition begun of consulting with faculty in other departments. Also, a charter was granted for the Utah Chapter of the ASA. Under Dale Richards’s chairmanship (1966–69), the faculty expanded from five to eight members. Gill Hilton, chair from 1969 to 1980, facilitated greater faculty research productivity and capitalized on this productivity by assisting the faculty in vending the state-of-the-art statistical software they developed.36
Al Rencher’s chairmanship (1980–84) helped the department acquire more physical space, additional faculty positions, and recognition for the consulting center. Under Lee Hendrix (1985–94) the undergraduate program expanded to join the largest in the nation and added actuarial science and quality science emphases. While Gale Rex Bryce was chair (1994–2000), enrollment in the introductory statistics class more than doubled to over four thousand per year; development of the introductory class along with additional positions to handle the increase became priorities. Under Howard Christensen, chair from 2000 to 2006, the use of multimedia and online technology was pioneered. The current chair (as of January 2012), Del Scott, has developed computing facilities and physical space, and both graduate and undergraduate curricula have undergone radical improvements.
All of the faculty now have PhDs in statistics or biostatistics, but in addition to developing statistical methodology, the faculty carry out research on a wide variety of applied problems, including air pollution monitoring, television ratings, improvement of sports teams, rating of sports teams, DNA analysis, health-care systems, evolutionary ecology, chemical thermodynamics, educational methods, industrial improvement, wildlife management, authorship styles in literature, paleontology, election prediction, and weapons systems reliability—just to name a few. The diversity has enriched each member of the department and led to beneficial cross-fertilization and collaboration.
A unique characteristic of the Statistics Department (at least among the physical and mathematical sciences) has been the relatively extensive involvement of women in the department almost from the start. Although the first graduating class consisted of five male students, most subsequent classes have included several female students. Among the first master’s students were several women.37 Even when there were no female faculty members in the department, a female graduate student was teaching introductory statistics in the department.38 As of 2012, the department has seventeen full-time faculty, four of whom are female (24 percent).
Fifty years after the Statistics Department slipped into existence, the department is strong and well poised for the future. The Jobs Rated Almanac has consistently rated both statistician and actuary among the top five careers for about twenty years.39 A 2009 editorial in the New York Times even reported the claim by the chief economist at Google that, given the staggering amount of data currently collected by business, government, and academia, “the sexy job in the next 10 years will be statisticians.”40 Fortunately, BYU is positioned advantageously to prepare LDS statisticians for these important future careers—all because of a whimsical comment by Howard Nielson more than fifty years ago and the dedication of those who followed him.
Sidebar: Howard C. Nielson
Born in Richfield, Utah, in 1924, Howard Nielson obtained a bachelor’s degree in mathematics from the University of Utah, a master’s degree from the University of Oregon in mathematics with a concentration in theoretical statistics, an MBA from Stanford, and a PhD in business with a minor in statistics from Stanford. After founding the Department of Statistics at BYU in 1960, Nielson became known among students for his quantitative abilities. John Lawson, a current member of the Statistics Department, recalled that Nielson could mentally do complicated arithmetic operations faster than students could do them with paper and pencil or the crude calculators of the time.
Nielson was able early on to establish a broad curriculum for the BYU department by, for example, staying a page or two ahead of the students in an operations research class and spending a summer term at the University of Wyoming learning design of experiments so that he could teach it the next fall. After leading the department for three years, Nielson was ready for new challenges. He took a sabbatical and turned the reigns of the department over to Mel Carter.
Although Nielson remained on the statistics faculty for nineteen more years, he was never again the dominant figure in the department. He returned from sabbatical to teach regular classes for two more years. In 1967, he won a seat in the Utah State Legislature, so he scheduled all of his classes for 6:00 and 7:00 a.m. His quantitative abilities landed him on the state budget committee as a freshman legislator. In 1970, he took a two-year leave from BYU to work for the Ford Foundation on Economic Development in Jordan. After the stint in Jordan, he returned to BYU and the state legislature, where he rose to the position of Speaker of the House.
In 1974, Nielson announced that he was running for the vacant United States Senate seat for Utah and again took leave from BYU. His wife’s health forced him to withdraw from the race, but by then the Department of Statistics had hired a replacement for him. He worked full time for Hercules Powder for a year and then, in a somewhat ironic shuffle, returned to BYU for a two-year position in the Economics Department as a temporary replacement for Merrill Bateman. He then spent four years as Utah’s Associate Commissioner of Higher Education, after which he successfully ran for a seat in the United States Congress in 1982. He served four terms as Congressman for Utah’s third district. At Nielson’s retirement party from the Statistics Department in 1982, Al Rencher made an only slightly exaggerated statement that Nielson had been a member of the department for twenty-two years, but had been on leave for twenty-four of them.
Nielson has 14 children from two marriages, 74 grandchildren, and 106 great grandchildren, at last count; a recent two-month window included two weddings, four births, and three baptisms. Since he retired, Nielson and his wife have served missions in Hungary and Australia.